1. Field of the Invention
The present invention relates to a highly efficient coding apparatus and, more particularly, to a highly efficient coding apparatus using predictive coding.
2. Description of the Prior Art
Prediction coding has been known as one highly efficient coding for lowering the bit rate of a digital picture signal. FIG. 1 shows its typical one. In FIG. 1, reference numeral 31 is a transmission side or encoder, reference numeral 32 is a reception side or encoder, and reference numeral 33 is a transmission line such as a digital line, wherein a process of magnetic recording/reproduction is provided between both sides
A digital picture signal from an input terminal 34 and a prediction value from a local decoder are given to a subtractor 35, and a difference signal from the subtractor 35 is supplied to a non-linear quantization circuit 36. The local decoder is composed of a representative value circuit 37, an adder 38, and a one sample delay circuit 39. The non-linear quantization valve circuit 36 and the representative circuit 37 have complementary characteristics with each other and are usually made up of an ROM.
The output data of the non linear quantization valve circuit 36 is given through the transmission line 33 to a representative circuit 41 on the reception side 32, and a quantization code is decoded into a representative value. The representative value is supplied to an adder 42 and added to the output signal of the adder 42 given through a one sample delay circuit 43. The output signal of the adder 42 is taken out at an output terminal 44 as a decoded output.
Such a highly efficient coding apparatus provides a high prediction efficiency and is an effective system when the sampling density is high. However, since a locally decoded value of a previous picture element is employed, there is a problem in the operation speed. Namely, the total delay amount of the subtractor 35, the non-linear quantization circuit 36, the representative value circuit 37 and the adder 38 must lie within one time slot of input data. For example, the prior art structure of FIG. 1 is not suited for the processing of a picture signal of the high bit rate such as a high vision signal.
For one method for solving such a problem, a system described "Adaptive Intraframe/Interframe Coding for HDTV signals by Using Extrapolative and Interpolative Prediction," an article published by the Institute of Electronics, Information and Communication Engineers of Japan, January 1987, Vol. J70-B No. 1, pp. 96-104 has been proposed.
In this method, picture elements are divided into two groups G1 and G2 to provide a quincunx. With respect to the G1 picture elements, extrapolative prediction using only the G1 picture elements, which have already been subjected to coding/decoding, is done in intraframe prediction and interframe prediction. As to the G2 picture elements, extrapolative prediction using the G2 picture elements in a previous frame, which have already been subjected to coding/decoding, is done in the interframe prediction, while with respect to the intraframe prediction, interpolative is prediction done by the G1 picture elements. With this method, since previous picture elements are not used in the intraframe extrapolative prediction, the processing time can be reduced. Also, as to the G2 picture elements, because only the G1 picture elements are used in the intraframe interpolative, no feedback loop is needed to allow the problem of the processing time to be solved.
However, in the apparatus described in the article, the interval between picture elements in the case of intraframe prediction is enlarged twice the original one with the result that correlation is lowered. As a result, there is a disadvantage that prediction efficiency is decreased.
Particularly, with increased prediction efficiency in the case of a fine pattern, there is a problem in picture quality in the worst case.
Further, there is considered a prediction coding method in which a picture is divided into plural ones and parallel prediction coding is done for every divided picture to improve the processing speed.
In this method, not only is there a problem that an initial value must be transmitted for each divided area but also there is a problem that the border of the areas is remarkably seen since its initial value portion must be processed exceptionally.